[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: virtual hosting



On Tue, 14 Jan 1997, Keith Moore wrote:
> What I can't figure out is why the submit server is more likely to 
> be able to generate correct messages than the UA. 

There are portions of the message headers which the submit server may know
more about.  The "date" header is an excellent example.  Most PCs don't
keep track of time in GMT and many have the wrong offset from local time
configured.

> If the MTAs leave a message unchanged, we have one source of error -
> the sender's UA.  If we introduce a submit server, we have two sources
> of error.  Either or both of these can be wrong, and the submit server
> doesn't necessarily have any better information than the UA.  What's
> more, the submit server doesn't know whether or not it has better
> information than a UA.

Any changes to the message by the submit server should be documented in
headers to help debugging.

> How is the submit server supposed to know who is actually sending
> the message?  If it doesn't know who the Sender is, it shouldn't guess.

In the presence of the SMTP AUTH extension, the submit server is likely to
have better knowledge of the Sender than the MUA.  It may be able to
verify the existance of a deliverable mailbox for the user.  MUAs running
on machines which don't also have the mailstore simply can't
generate a correct Sender header (which is why I wrote the originator-info
internet draft).

> If my local users insisted on using broken UAs couldn't be fixed, I'd 
> certainly consider routing their outgoing mail through a filter that would
> attempt to fix up the message.  But it's a stopgap measure, to be applied 
> only in specific cases to solve specific problems.  It's not appropriate 
> for widespread use or standardization.  

I agree.  For any fix which meets the following two criteria:

(1) The MUA knows how to generate the information better than the
submission server.
(2) The MUA format is not valid RFC 821/822.

The submission server should either reject the message or send a trouble
report back to the sender after fixing it.  It'd be good to make the rules
for the submission server good enough so that MUA writers who support it
will be told about their screwups by the server.